Prebuilt VM images for JAX on GCP

[AakashKumarNain]

I was trying to do install JAX with GPU support on a linux machine. I followed the instructions provided here but conda install complains about the glibc version. Upgrading glibc can break a lot of things on the system.

Given that there are VMs for TF, and PyTorch on GCP, are there any plans to provide VMs for JAX?

[ayaka14732]

To address your concern, it might be more beneficial to use pip for installation rather than conda. Pip, as the Python package manager, is typically more suited for managing Python packages.

Opting for a virtual machine might not be the optimal solution. It would be suggested to follow the installation guide provided on the JAX GitHub page: https://github.com/google/jax#installation. This guide allows the installation of the latest version of JAX using pip. After setting the version, the same command can be used on other machines to ensure version consistency.

While the convenience of prepackaged tools or images can be appealing, it might also introduce additional complications. Understanding and learning the correct usage of these tools could greatly enhance problem-solving efficiency in similar situations in the future.